Automatically suggesting regions for blur kernel estimation

Abstract

A computer-implemented method and apparatus are described for automatically selecting a region in a blurred image for blur kernel estimation. The method may include accessing a blurred image and defining a size for each of a plurality of regions in the image. Thereafter, metrics for at least two of the plurality of regions are determined, wherein the metrics are based on a number of edge orientations within each region. A region is selected from the plurality of regions based on the determined metrics, and a blur kernel for deblurring the blurred image is then estimated for the selected region. The blurred image is then deblurred using the blur kernel.

What is claimed is: 
     
       1. A computer-implemented method, comprising:
 accessing a blurred image; 
 defining a size for each of a plurality of regions in the blurred image; 
 determining metrics for at least two of the plurality of regions, the metrics being based on a number of edge orientations within each region, a gradient magnitude of pixels within the region, and a weight associated with the pixels within the region; 
 selecting a region from the plurality of regions based on the determined metrics; 
 based on the selected region, estimating a blur kernel for deblurring the blurred image; and 
 deblurring the blurred image using the blur kernel to produce a deblurred image. 
 
     
     
       2. The method of  claim 1 , wherein selecting the region is based on the number of edge orientations of the region exceeding a threshold value. 
     
     
       3. The method of  claim 2 , wherein the metrics are further based on a usefulness factor, the usefulness factor determined from the number of edge orientations and an image gradient. 
     
     
       4. The method of  claim 3 , wherein the metrics are further based on a number of over-exposed or under-exposed pixels, an influence of the over-exposed or under-exposed pixels being weighted. 
     
     
       5. The method of  claim 1 , wherein the metrics are further based on a location weight associated with a location of the region within the blurred image. 
     
     
       6. The method of  claim 5 , wherein the location weight is a pixel weight function determined at least partially by a distance between a center pixel in a region of the plurality of regions and a center of the blurred image. 
     
     
       7. The method of  claim 6 , further comprising:
 receiving a user input identifying a user defined location at which to estimate the blur kernel in the blurred image; and 
 modifying the location weight based on the user input. 
 
     
     
       8. The method of  claim 1 , wherein the blurred image is downsampled with respect to the size of the blur kernel. 
     
     
       9. The method of  claim 1 , further comprising automatically, without user input, determining the size of the blur kernel. 
     
     
       10. The method of  claim 1 , the method further comprising:
 modifying a size of each of the plurality of regions; 
 determining metrics for each of the plurality of regions having a modified size; and 
 selecting a region having the modified size that is associated with a metric that satisfies a threshold metric for estimating the blur kernel. 
 
     
     
       11. An image deblur system, comprising:
 one or more processors; 
 memory, coupled with the one or more processors, having instructions stored thereon, the instructions, when executed by the one or more processors, to cause the image deblur system to: 
 access a blurred image; 
 determine metrics for at least two regions of a plurality of regions, the metrics being based on a number of edge orientations within each region, a gradient magnitude of pixels within the region, and a weight associated with the pixels within the region; 
 to select a region of the plurality of regions based on the determined metrics; and 
 to estimate a blur kernel for deblurring the blurred image, wherein the blur kernel is based on the selected region. 
 
     
     
       12. The system of  claim 11 , wherein to select the region is based the number of edge orientations of the region exceeding a threshold value. 
     
     
       13. The system of  claim 11 , wherein the metrics are further based on a usefulness factor, the usefulness factor determined from the number of edge orientations and an image gradient. 
     
     
       14. The system of  claim 13 , wherein the metrics are further based on a number of over-exposed or under-exposed pixels, an influence of the over-exposed or under-exposed pixels being weighted. 
     
     
       15. The system of  claim 11 , wherein the metrics are further based on a location weight associated with a location of the region within the blurred image. 
     
     
       16. The system of  claim 11 , wherein the instructions further cause the system is configured to:
 modify a size of each of the plurality of regions; 
 determine metrics for each of the plurality of regions having a modified size; and 
 select a region having the modified size that is associated with a metric that satisfies a threshold metric for estimating the blur kernel. 
 
     
     
       17. A computer-readable storage device including instructions which, when executed by a computer, cause the computer to perform operations comprising:
 accessing a blurred image; 
 defining a size for each of a plurality of regions in the blurred image; 
 determining metrics for at least two regions of the plurality of regions, the metrics based on a number of edge orientations within each region, a gradient magnitude of pixels within each region, and a weight associated with the pixels within each region; 
 selecting a region from the plurality of regions based on the determined metrics; 
 based on the selected region, estimating a blur kernel for deblurring the blurred image; and 
 deblurring the blurred image using the blur kernel to produce a deblurred image. 
 
     
     
       18. The computer-readable storage device of  claim 17 , the region is selected based on the number of edge orientations of the region exceeding a threshold value.